Live virus vaccines offer some of the biggest successes of medicine. They offer superior immunogenicity over inactivated virus vaccines, but they have two drawbacks. First, methods for creating attenuated vaccines are hit-and-miss. Second, successful live-virus vaccines can often evolve back to high virulence. Indeed, polio eradication has remained elusive because of vaccine evolution. The first of these hurdles is potentially surmountable with genome engineering, if we can predict the viral fitness effect of the engineering. The second hurdle may also be overcome by engineering if we understand the molecular evolution of engineered viruses. This proposal develops a combined empirical and computational viral system to study attenuation of evolutionary reversal of that attenuation. The virus is a dsDNA bacteriophage (T7) that is safe, easily manipulated and engineered. With its extensive background of genetic, biochemical and evolutionary studies, it offers the best empirical and theoretical foundation of all viruses for addressing this problem. Our approach consists of three Aims that collectively combine genome engineering with molecular studies of the viral life cycle, fitness measures, evolution of attenuated genomes, sequence analysis, and computational modeling. In Aim 1, we build several genomes to test new methods of viral attenuation: silent codon modification, genome rearrangement, and promoter deletion. Two questions motivating this work are (i) whether the level of attenuation is predictable, and (ii) whether the attenuation is evolutionarily stable against reversion to high fit- ness. Beyond genome construction, we will thus measure fitness of all constructs and evolve all constructs for hundreds to thousands of generations, observing fitness recovery and sequence evolution. This Aim stems from a variety of preliminary work demonstrating feasibility of all technical aspects and is the foundation of Aims 2 and 3. Aim 2 is the application of key molecular assays to the viruses from Aim 1, proteomics, transcriptomics, and RNA densities on ribosomes (ribosomal profiling). The intracellular viral life cycle will be described at the level of transcription, translation, and protein abundance to understand the molecular bases of different attenuation methods and the paths of evolutionary recovery. Initially, we will compare patterns transcript and protein abundances with ribosome densities on mRNAs to determine whether they agree with each other and match expectations from the basic biology of the virus. Unexpected patterns will be confirmed experimentally. These methods will provide insight to how the different engineering methods attenuate and how they retard evolutionary recovery. Further, these methods will be used to parameterize and further develop an existing virtual model that gives an overall predictive framework for attenuation and evolution (Aim 3). Aim 3 consists of modeling and analysis. Initially, an existing, second-generation computational model of the viral life cycle will be calibrated to the data we obtain for wt and attenuated viruses. As the model incorporates transcription and translation, the molecular data obtained will be compared at a mechanistic level to model predictions. We expect that translation is modeled with insufficient detail in the current model, and we will develop a third-generation model that addresses the shortcomings of the current model. Ultimately, the model we are developing will be useful for prediction of both attenuation and evolution of T7.